noyb Accuses Scale AI of Watching Its AI Trainers Keystroke by Keystroke
Privacy group noyb has filed five GDPR complaints alleging Scale AI tracks and face-scans the 900,000 freelancers on its Outlier platform. Scale did not immediately respond to Reuters, and no regulator has ruled.
The hidden workforce that teaches AI models to behave is being watched harder than most office staff, according to a new set of privacy complaints. European privacy group noyb says Scale AI tracks the freelancers on its Outlier platform with up to 19 trackers, records their keystrokes and puts them through repeated face scans. Those are noyb's allegations. They are not findings by any regulator.
On Tuesday noyb filed five GDPR complaints against Scale AI, according to its own announcement. Two went to Austria, and one each to France, Italy and the Netherlands, per Euronews. The sources I read do not name the specific data protection authorities that received them. It filed with Interet a Agir, DiPLab and Reversing.Works.
Outlier is the target because of its scale. The platform has more than 900,000 freelancers who write prompts, rank answers and correct model output. Reuters reported that their recorded tasks become training data for clients such as Meta, OpenAI and Google. Meta owns 49 per cent of Scale AI. noyb also lists xAI, Anthropic and the US government among customers. Being a customer is not an accusation, and none is alleged to have ordered the tracking.
The surveillance claims come in layers. According to Reuters, the complaints say up to 19 trackers follow workers, recording keystrokes and mouse movements every few seconds. A DiPLab investigation found the resulting file could be rebuilt into a video that replays a work session from start to finish.
noyb's own technical analysis adds a second claim. It says trackers from TikTok, LinkedIn, Facebook, the survey platform Sprig and the data analysis company Datadog land on workers' devices while they are on Outlier. Every few seconds, noyb says, they send out IP addresses, contact details, unique identifiers and the page being visited, data that can feed targeted advertising and profiling.
"Why should LinkedIn or Facebook know which tasks you're working on?" Lisa Steinfeld Leiser, data protection lawyer at noyb
Then there is the face check. Workers must repeatedly submit ID documents and video of their face to the verification firm Persona. noyb says Scale AI justifies this as fraud prevention and a way to verify skills, but argues it goes far beyond what is necessary. The complaints, per Reuters, call the checks invasive, repeated without explanation and contrary to EU biometric rules. noyb also says about 70,000 photos of government issued IDs stored with Persona were exposed by hackers in October 2025.
The complaints also cover how work is handed out. Outlier uses game-like mechanisms, and bonuses are central to its incentives, DiPLab found. Researcher Thomas Le Bonniec said Outlier offered $100 to $200 for completing 200 tasks in 24 hours, citing worker interviews. He added that Italy's data protection authority fined delivery platform Glovo in 2021 over similar gamification.
"You don't work when you want. You work when the platform wants you to and there are tasks available." Thomas Le Bonniec, DiPLab researcher
One allegation is simple enough to check without a technical audit. Four workers asked Outlier for copies of their data between December 2025 and May 2026, and none got an answer, the complaints say. noyb argues that breaches the GDPR right of access. It also says Scale AI has no valid legal basis to place trackers without consent and violates the principle of data minimisation.
What noyb wants is a full investigation, and it suggests the authorities impose an administrative fine. Reuters adds that the group wants regulators to examine how the platform assigns tasks. Nothing is decided. Scale did not immediately respond to Reuters' request for comment,.
The bigger point is uncomfortable for the whole industry. Chatbots sell as clean software, yet noyb says they lean on hundreds of thousands of people working under conditions that would raise eyebrows in a normal European job. Baptiste Delmas of Interet a Agir put it bluntly, saying the AI boom relies on a vast workforce whose privacy is at stake. Whether regulators agree is the question now.
